A Python toolbox for analysing large music collections and long music recordings.
Point it at a folder of audio files, where each subfolder counts as an album, and it renders figures, tables and thumbnails that let you compare many tracks at a glance: albums, similarity, outliers, categories. Point it at a concert recording instead and it finds the songs inside it first, so the same tools apply to a live set.
pip install musiscapemusiscape report ~/Music/my-collectionOne command runs the whole pipeline and writes analysis/README.md next to
your music: an album table, a similarity landscape, self-explaining
categories, and the corpus extremes.
| Command | What it does |
|---|---|
probe |
list albums and tracks, no analysis |
extract |
per-track features, cached in features.json |
fingerprint |
per-album profile bars |
landscape |
PCA similarity map and album-affinity matrix |
categorize |
k-means categories with named signatures |
report |
everything above, gathered into analysis/README.md |
thumbnails |
one visual card per track (--style, seventeen styles) |
poster |
the whole collection as one image (--style vinyl for discs) |
sonic |
a ~12-second audio summary per track, plus album medleys |
segment |
finds the songs in a concert recording, labels every second of it, and writes one file per song |
figures |
labelled chromagram and tempogram per track, at any pixel width |
pdf |
one PDF: a summary table, then a page of figures per track |
A live recording is not a collection until it is cut into songs, which is what segment is for
(video files included, since concerts usually arrive as video):
musiscape segment ~/video/concert -o ~/video/concert/analysis
musiscape report ~/video/concert/analysis/songsCommon options: -o sets the output folder (default <root>/analysis),
--workers N parallelises, --duration S analyses only the first S seconds
per track, and -k fixes the number of categories.
- Documentation—guides, an illustrated gallery of the visualisation styles, and the API reference
- Wiki—methodology notes and a case study
The toolbox is meant for quick visualisations and overviews. Combine it with listening.
These toolboxes come out of the fourMs lab at the University of Oslo.
- Musical Gestures Toolbox
(
musicalgestures): video and audio, with motiongrams, videograms and motion analysis from ordinary video files - ambiscape: soundscapes, the sonic ambience of a place, including multichannel and ambisonic material
- micromotion: human micromotion, with quantity of motion from optical markers, accelerometers, respiration belts and force plates
Each is a separate package with its own release cycle, and each runs on its own. musiscape needs none of the others installed to do anything it documents.
They are still meant to be combined, so any measure that appears in more than one of them has a single owner and a single implementation. Circular statistics belong to micromotion, and musiscape imports pulse clarity, fifths-circle centres and the Rayleigh test from there rather than keeping its own copy. Switching a measure between toolboxes, or computing it in two of them, gives the same number. A test suite in each package checks that against the owning implementation, and skips when the sibling is not installed.
MIT licence. musiscape is developed as part of the AMBIENT project at RITMO Centre for Interdisciplinary Studies in Rhythm, Time and Motion, University of Oslo.
Cite the concept DOI, which always resolves to the newest version:
Jensenius, A. R. (2026). musiscape: A Python toolbox for analysing large music collections and long music recordings [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.21964192
Where the exact behaviour matters, add the version you ran. Every release has its own DOI, listed on the Zenodo record.
CITATION.cff in this repository carries the same information in machine-readable form.